Serotype K1 Capsule, Rather thanmagAPer Se, Is Really the Virulence Factor inKlebsiella pneumoniaeStrains That Cause Primary Pyogenic Liver Abscess

To the Editor —In the 1 March 2006 issue of the Journal Chuang et al. described the complete sequence of the magA flanking region [1]. We have performed a similar study [2]. By serotyping the original strain harboring magA and comparing it with cps (capsular polysaccharide synthesis) loci in both Klebsiella pneumoniae serotype K2 and K52, Chuang et al. confirmed that magA which is located within the cps gene cluster, is responsible for the synthesis of serotype K1 capsule. After screening 74 strains, of which only 40 were typeable and were restricted to 4 serotypes, they found that all serotype K1 strains were magA positive and that all strains of the other 3 serotypes were magA negative. In addition, 35 of 42 strains from patients with primary pyogenic liver abscess (PLA) were serotype K1/magA positive. Furthermore, Chuang et al. identified 5 more open reading frames (ORFs) in the magA flanking region that shared the same feature. Thus, they asserted that these 6 ORFs, because they are unique to K. pneumoniae serotype K1, could be used for rapid molecular typing and early detection of PLA. However, we consider the evidence that supports that claim to be weak, given that the screening included strains from only 4 serotypes and that there are at least 77 serotypes of K. pneumoniae; there is the possibility that the above-described ORFs exist in the cps gene clusters of serotypes that were not tested In Fang et al.’s study of magA this same group of investigators searched for virulence genes in K. pneumoniae strains that cause PLA by mini-Tn5 transposon mutagenesis [3]. They obtained 2500 mutants from a PLA strain and selected 20 “less virulent” mutants. However, they chose less mucoid mutants either on the basis of gross appearance or by the string test but did not assay directly by any virulence test, such as the serum-resistance assay or the phagocytosis assay. In light of this selection bias, we think that they have identified the gene that influences capsule synthesis rather than the gene that determines virulence Among the mutant genes, many of which were identified as part of the cps gene cluster, Fang et al. found a 1.2-kb ORF that fulfilled the molecular version of Koch’s postulates and that was more prevalent in PLA strains. Because this ORF did not exist in the cps gene cluster of serotype K2 and the genome of MGH-78578 (K52), they considered it to be a novel virulence gene and named it magA (mucoviscosity-associated gene A). The sequence of the magA flanking region in Chuang et al.’s study was previously reported by the same group [3] Fang et al. had a collection of 53 PLA strains, of which 52 were magA positive. They explained that the 1 PLA strain that was magAnegative resulted from the patient’s poor medical condition (liver cirrhosis and hepatic failure), and they created the image of a 100% rate of detection of magA in PLA strains. Our principal objection is that, even with a similar period of strain collection done by the same group, there is a significant difference in magA prevalence in PLA strains between Fang et al.’s study and Chuang et al.’s study (52/53 vs. 35/42; P<.01). Whether a selection bias existed in their studies should be clarified In a previous surveillance study, we reported that the prevalence of serotype K1 among K. pneumoniae PLA strains was 63.4% [4]. Consequently, we screened all of our 134 PLA strains and found that all of the K1 strains were magA positive and that all of the non-K1 strains were magA negative [2]. In a more extensive survey using a collection of 495 isolates encompassing the reference strains of all 77 serotypes, Struve et al. showed that magA is restricted to the cps gene cluster of K. pneumoniae serotype K1 [5] The sensitivity of the use of magA in the detection of PLA depends on the prevalence of serotype K1 among K. pneumoniae strains that cause this disease. A K1 prevalence rate of no more than 65% among K. pneumoniae PLA strains has been reported [5–8], except for the rates reported by Fang et al. and Chuang et al. If the actual prevalence of serotype K1 is not as high as they report, magA may be not a good marker for the detection of PLA On the basis of the evidence of the location and serotype specificity of magA we deduce that the loss of virulence and mucoviscosity of the magA− mutant may be attributed to the disruption of the serotype K1 capsule. Thus, the serotype K1 capsule, rather than magA per se, is really the virulence factor in K. pneumoniae PLA strains